Monitoring the Electrical Behaviour of the Electrode-Tissue Interface by way of Reverse Telemetry in a 100 Channel Neurostimulator

Size: px

Start display at page:

Download "Monitoring the Electrical Behaviour of the Electrode-Tissue Interface by way of Reverse Telemetry in a 100 Channel Neurostimulator"

Hugh Barker
5 years ago
Views:

1 Monitoring the Electrical Behaviour of the Electrode-Tissue Interface by way of Reverse Telemetry in a 100 Channel Neurostimulator Gregg J. Suaning* Ψ, Wayne L. Gill Ψ, Nigel H. Lovell Ξ Ψ - University of Newcastle, Australia, Ξ - University of New South Wales, Australia Introduction Knowledge of the state of the electrode-tissue interface in Functional Electrical Stimulation (FES) is an important factor in evaluating the chronic viability of such applications. Changes in the electrical impedance of a given electrode-tissue interface can be attributed to several factors including mechanical movement, electrode dissolution, changes in tissue morphology, and device degradation or failure. It is known from other applications that stimulation thresholds change over time [1], and while patient adaptation appears to play an important role in these changes, the influence the state of the electrode-tissue interface has on stimulation thresholds is not fully understood. As more sophisticated neuroprostheses are developed and the quantities of stimulating sites increase, the need arises for means of monitoring the electrical characteristics of the electrode-tissue interface through non-invasive means. Previous reports from the first and last authors have described the neuroprosthesis used in the experimentation described herein [2]. The present paper aims to characterise the capabilities of the device s reverse telemetry facilities in predicting the end-of-phase voltage across any electrode combination, so as to provide a means for in-situ monitoring of some aspects of the electrical characteristics of the electrode-tissue interface over time. (a) Figure 1 (a) Theoretical voltage waveform from a constant current source - as measured across a pair of biological electrodes. Subscript dl denotes double-layer capacitance at the electrode surfaces (electrodes 1 and 2). IR is the voltage that must be placed across the electrodes prior to any charge transfer occurring. The pulse width is denoted by the parameter t. (b) Proposed 100 site nerve cuff electrode comprising silicone elastomer substrate with cavities for 100, 450um diameter, spherical Pt electrodes. (b)

2 Methods The application specific integrated circuit (ASIC) that serves to drive the implantable stimulator possesses an intrinsic circuit that stores the end-of-phase voltage across the electrodes through which constant current stimulus has passed. Upon external request from the body-worn hardware that powers and controls the stimulator by way of radio-frequency (RF) commands, the implanted device subsequently sends a temporallyencoded signal corresponding to the end-of-phase voltage. This signal (henceforth known as the telemetry signal) is detectable via inductive coupling with the implanted device using hardware external to the body. As such, the end-of-phase voltage may be acquired via the telemetry signal (following digitisation and processing) in a non-invasive fashion following implantation. The capture of the telemetry signal requires amplification of a small (tens of microvolt order) and brief (tens of nanosecond order) signal acquired from an inductor placed in the vicinity of the implant site, but outside the body. The time at which the telemetry signal arrives coincides with a period wherein RF is not being sent to the implant so as to make possible the signal s detection. Knowledge of the timing of the stimulus phases, a logic signal readily accessible on the external hardware that controls the implant, allows for relatively straightforward recording of the timing between the end of stimulus, and the arrival of the telemetry signal. Characterisation of the end-of-phase/telemetry signal relation was achieved by stimulating through a series of fixed resistors with negligible capacitance in comparison to the capacitive effects present with electrodes within electrolyte. In the absence of capacitive effects, the end-of-phase voltage is independent of pulse width. Characterisation data was acquired for each fixed resistor for each of the 31 current levels in the five-bit digital to analogue converter (DAC) present on the ASIC. For each DAC setting, cubic spline interpolation equations were generated so that prediction of the end-of-phase voltage for any telemetry signal detection time could be made. These interpolation equations were coded directly into the telemetry signal acquisition software for immediate conversion of telemetry signal detection to end-of-phase voltage. Following generation of the cubic spline interpolation equations, 450 micrometre diameter, pure platinum, spherical electrodes consistent with those proposed for the FES cuff electrode illustrated in Fig. 1(b) were fabricated using high-energy electrical discharge [4]. These electrodes were insulated to the base of the sphere and placed in physiological saline solution where subsequent biphasic stimulation was delivered for a series of pulse widths (1, 5, 10, 50 times 25.6 microsecond duration) for all 31 DAC settings ranging from to 1.8 ma. For each set of stimulus parameters (pulse width, DAC setting), the telemetry signal was recorded and converted to end-of-phase voltage via the cubic spline interpolation equations. At the same time, a digital storage oscilloscope (DSO, Tektronics TDS220) was connected to the electrodes (outside the saline bath) such that the actual end-of-phase voltage could be measured. It must be noted that the DSO did make a small but apparent influence to the capacitance and impedance of the stimulation circuit. As such, the stimulation circuit was effectively treated as a black box with the electrodes in saline in parallel with the DSO comprising the black box s contents. Without the DSO in place, the comparison of measured and predicted end-of-phase voltages could not be made. Results Fig. 2(a) shows the measured end-of-phase voltage in comparison with the predicted end-of-phase voltage from the cubic spline interpolation equations for the example of DAC setting 15 of 31 (1.1 ma). The derivative of the trace shown in Fig. 2(b) illustrates the end-of-phase voltage/telemetry signal sensitivity for this DAC setting, an important factor in selecting the most appropriate DAC setting for the telemetry measurement across a given electrode configuration. In this example, voltages below approximately 250mV cannot be accurately predicted owing to small variations in the acquired telemetry data leading to large variation in the predicted end-of-phase voltage.

3 (a) Figure 2 (a) Comparison of measured and spline-fit predictions of end-of-phase voltage for the charge recovery (anodic) phase of stimulation for DAC setting 15 (1.1 ma) across two, 450 um diameter, Pt sphere electrodes in physiological saline. (b) Calibration data for DAC setting 15. It is from these datapoints that the cubic spline interpolation equations are derived. (b) Discussion The correlation of predicted and measured end-of-phase voltages illustrated in Fig. 2(a) indicates that telemetry provides a viable means of predicting the end-of-phase voltage in an implant scenario wherein no other accurate means of determining these figures are possible without invasive measures. The value of this information, however, requires some interpretation noting that the data presented in Fig. 2(a) is not the same as what one would observe when viewing the stimulus waveform of 1.25 ms duration on a DSO. The voltage-time waveform shown in Fig. 1(a) illustrates the capacitive charging as a consequence of constant current injection into the so-called double-layer capacitance (C dl ) formed at the electrode-tissue, or, more precisely, the electrode-electrolyte interface. The magnitude of C dl is a function of a number of factors including electrode surface area, surface materials, the electrolytic properties of the tissue medium, etc. Readers familiar with the actual voltage-time waveform of stimuli will recognise that the linearity of the waveform shown in Fig. 1(a) is theoretical and Faradic charge injection occurs in most practical applications. Monitoring of the true voltage-time shape, inclusive of non-linearity, of the biphasic stimulus waveform may also be achieved by way of the telemetry signal, but not without knowledge of the form of the first phase of stimulus. As described above, the telemetry signal indicates the end-of-phase voltage for the second phase of stimulus. To acquire information on the first phase of stimulus, one must trick the implant into not sending the first phase of stimulus, therefore ensuring the only phase is the second phase and the telemetry data acquired under these conditions will pertain to monophasic stimulus alone. This, in theory and in practise, will yield data on the first phase of stimulus of a biphasic waveform as future events (the second phase) cannot influence the past (the first phase). It should be noted here that the telemetry delivers the magnitude of the end-of-phase voltage so whether or not the phase is anodic or cathodic is of no consequence.

4 For monophasic stimuli, successive acquisition of the telemetry signal with incremental changes in the pulse width (t in Fig. 1(a)) would, at first glance, directly facilitate the reconstruction of the monophasic stimulus (voltage) waveform. However, residual charge on the electrode surfaces following each stimulus event will serve to offset the voltage waveform at the onset of subsequent stimulations without the presence of further forms of charge recovery or dissipation. Accordingly, monophasic stimuli has been found to contribute to irreversible tissue damage and thus a charge-balanced biphasic waveform proposed by Lilly et al. [3] is delivered in preference. The presence of the first (cathodic) phase of stimuli serves to offset the voltage waveform of the second (anodic) phase and it is this offset that substantially influences the results shown in Fig. 2(a). In other words, while the end-of-phase voltage is accurately portrayed in Fig. 2(a) for each time interval when data was acquired, the voltage waveform is not being reconstructed owing to the offset present at the beginning of the second phase of each stimulus pulse and this offset is pulse width dependent (refer to the final voltage shown at the end of stimulation in Fig. 1(a) it is non-zero, and pulse width dependent). If the ultimate objective were to be biphasic waveform reconstruction, the telemetry acquisition method described herein would appear to facilitate this, albeit in an indirect way. First, one must establish the shape of the monophasic (first phase) voltage waveform. As such, the implant must be commanded to by-pass the first phase of biphasic stimulation and send only monophasic stimuli. While the safety protocol intrinsic to the implant prevents this from occurring, it does, however, allow for a very brief (microsecond order) first phase that would produce insignificant charge injection relative to the much larger second phase. One would be effectively tricking the implant in this way, satisfying the safety protocol by ensuring that a first phase exists, but making it so small as to be of no significant consequence in terms of charge injection. Following a monophasic stimulus, the electrodes would be shorted together so as to negate charge imbalances between electrodes caused by the charge injection during stimulation that is not recovered in the absence of biphasic stimuli. This occurs at the expense of the time that is necessary for the charge dissipation to take place this is impractical where highspeed stimulation is concerned, but of little consequence when performed occasionally for purposes of research, diagnostics or measurement. Thus, monophasic stimulus pulses of various durations may be delivered, briefly and thus harmlessly, with long periods of electrode shorting between them. Owing to the shorting and charge dissipation, each event of monopolar stimulation begins from the same base-line voltage. With each event of stimulation, telemetry is used to predict the form of the monophasic stimulus voltage as a function of time by way of variation of pulse duration. The voltage IR shown in Fig. 1(a) represents the nearly pure resistive contribution of the electrolyte. An estimate of this voltage may be obtained by acquiring telemetry for a monophasic burst of very short duration, or, perhaps more accurately, extrapolating the telemetry data to predict the voltage at the beginning of the phase. Ignoring non-linearities, the progression of a biphasic waveform exists as follows: (i) from the baseline voltage, the cathodic phase begins with an abrupt negative voltage of magnitude IR (ii) as current begins to flow, the double-layer capacitance begins to charge (negatively) and continues to charge until the end of the phase (iii) an abrupt rise in voltage of magnitude IR occurs to complete the phase. Telemetry, it must be noted, measures the end-of-phase voltage before the IR rise, thus yielding the peak voltage magnitude during stimulus. (iv) a brief inter-phase gap takes place during which no substantial voltage change occurs (v) the anodic phase begins with an abrupt positive voltage of magnitude IR note the 2 x IR rise in voltage since the peak magnitude of phase one was acquired (vi) the double-layer capacitance begins to charge (positively) and continues to charge until the end of the phase (vii) an abrupt drop in voltage of magnitude IR occurs to complete the phase, again with telemetry yielding the peak voltage prior to this drop. For a constant current, charge balanced, biphasic waveform, the voltage across the electrodes should now have returned to the original baseline. In practise, this is not precisely the case and subsequent charge recovery is often warranted albeit on a far smaller scale than is required for monophasic stimuli. Establishment of the shape of the monophasic waveform and the magnitude of the voltage IR will establish the starting voltage for the second phase when biphasic stimulus is subsequently delivered. With predictions of the

5 starting voltage for each pulse width now known, the ASIC would return to normal operation, producing a charge balanced, biphasic waveform with each stimulation event. With each of these, the end-of-phase voltage is predicted using telemetry. By subtracting the starting voltage of the second phase of stimulus from the predicted voltage at each measured pulse width interval, prediction of the actual waveform would eventuate. An efficient and automated method of constructing the actual biphasic voltage waveform is work currently underway by the authors. Conclusions Using only telemetry data, one may accurately acquire the end-of-phase voltage for a given stimulus waveform. In a clinical scenario, this information may be used to monitor changes over time of the physical electrode-tissue interface, and serve as a means of diagnosis of change, damage or failure in the stimulating device. While theoretically possible, reconstruction of the actual voltage-time waveform of the stimulus, as one would obtain from direct monitoring with a DSO for example, is the topic of ongoing pursuit by the authors. References [1] Pfingst BE (1990) Changes over time in thresholds for electrical stimulation of the cochlea. Hear Res, 50: [2] Suaning, G.J., Lovell, N.H. (2001) CMOS neurostimulation system with 100 channels, scalable output and bi-directional radio frequency telemetry. IEEE Trans. Biomed. Eng., 48(2); [3] Lilly JC, Hughes JR, Alvord EC Jr., Galkin TW. (1955) Brief, noninjurious electric waveform for stimulation of the brain. Science, 121:468 [4] Suaning, GJ. Lovell, NH. Kwok, CY. (2002) Fabrication of platinum spherical electrodes in an intra ocular prosthesis using high energy electrical discharge, Pacific Rim Workshop on Transducer and Micro/Nano Technologies, July 22-24, Xiamen, China Acknowledgments: The authors wish to acknowledge the Australian Research Council for their support of this research.

Neural Stimulation with Active Charge Balancing Feng Wang, Phuc-linh Nguyen, Jonathan Helm, Jimmy Zong

Neural Stimulation with Active Charge Balancing Feng Wang, Phuc-linh Nguyen, Jonathan Helm, Jimmy Zong Introduction We propose to design a micro-stimulation circuit cell for use in visual prosthesis applications.